3.2.1
General Characteristics
GMOS-N is a multi-object spectrograph on the 8 m Gemini North telescope which was built by a collaboration between UK and Canada, was delivered in July 2001 and entered regular science service in late 2001. Three 2048 × 4608 CCD with 13.5 µm pixel, optimised for the red end of the optical length region, are used to provide 0.36-1.1 µm long-slit, multi-slit spectroscopy broad and narrow-band imaging over a 5.5 arcmin field of view [150]. Spectral resolution varies with different gratings (Table 3.1). A summary of the main instrument specifications is given in Table 3.2 and a schematic of the detector array is shown in Figure 3.1.
3.2.2
MOS
In multi-slit mode GMOS-N is able to locate several hundred slits in a single mask. A total of 18 masks can be loaded into GMOS at any given time with up to six
3.2 The Gemini Multi-Object Spectrograph North 95
Characteristic Specification
Wavelength Range 0.36-1.1 µm
Field of view 5.5 × 5.5 arcmin2
Image scale at detector 0.07200per 13.5 µm pixel
Minimum slit width 0.200
Spectral resolution R = 5000 to R =500 with 0.200slit
Integral field Capability Remotely deployable, 0.200pixels with 52 arcsec2 field
Detector 6144 × 4608, 13.5µm pixel 3 × 1 EEV CCD array
Table 3.2: Specifications of the GMOS-N instrument [150].
2004 PASP, 116:425–440
TABLE 1
Specifications of the GMOS-N Instrument (as Built)
Characteristic Specification
Wavelength range . . . 0.35–1.1 mm Field of view . . . 5!.5 # 5!.5
Image scale at detector . . . 0".072 per 13.5 mm pixel
Minimum slit width . . . 0".2
Spectral resolution . . . R p 5000 toR p 500 with 0".5 slit
Integral field capability . . . Remotely deployable, 0".2 pixels with 52 arcsec2 field
Detector . . . 6144# 4608, 13.5 mm pixel 3# 1 EEV CCD array
Fig. 1.—Schematic diagram showing the layout of the three GMOS detectors
that form the 6144# 4608 pixel array. There are small gaps between the
detectors of about 0.5 mm, corresponding to about 37 pixels. The imaging field of view occupies the central region of the array and is shown by the shaded region. The on-instrument wave-front sensor patrol field, projected onto the detector plane, is shown by the dotted line. Note that there is a reflection in the vertical direction between the detector plane shown here and the mask plane shown in Fig. 3 of Murowinski et al. (2004). In spectroscopic mode, a slit mask is moved into the beam to cover the imaging field, and the resulting spectra run horizontally across the CCD array. In the long-slit case, the slit runs vertically up the center of the field.
After installation on the Gemini–North telescope in 2001
August, GMOS commissioning began with daytime work
(characterization of flexure, etc.), followed by nighttime tests.
In this paper, the main results are summarized from the night-
time commissioning of the imaging, long-slit, and MOS modes.
First results from the IFU commissioning have been presented
by Allington-Smith et al. (2002).
In a separate paper (Murowinski et al. 2004), details are
presented of the instrument design and the extensive laboratory
tests that the instrument underwent before shipping. Compre-
hensive overviews of the GMOS designs, including novel fea-
tures such as the OIWFS and flexure compensation system,
have been presented elsewhere (Davies et al. 1996; Murowinski
et al 1998, 2004; Crampton et al. 2000). A summary of the
main instrument specifications is given in Table 1, and a sche-
matic diagram of the detector array is shown in Figure 1.
In the following sections we describe nighttime observations
that were used to test the on-sky performance of GMOS in
imaging, long-slit, and multiobject spectroscopic modes.
2. THROUGHPUT MEASUREMENTS
GMOS was designed to be a high-throughput spectrograph
and makes use of special optical coatings and glasses (including
large calcium fluoride lenses) to meet this goal in addition to
tight specifications on image quality (Stilburn 2000; Murow-
inski et al. 2003b). Laboratory measurements of the throughput
of individual components that make up GMOS, including the
EEV CCDs, are shown in Figure 2. In this section, the expected
response of GMOS based on these response curves is compared
to the throughput measured from observations of standard stars.
2.1. Throughput in Imaging Mode
The throughput of the GMOS-Gemini–North system in im-
aging mode was measured from observations of the standard
star field PG 1323-086 (Landolt 1992), observed on 2003 Feb-
ruary 5. GMOS was mounted on one of the side-looking ports
of Gemini–North during these observations. The data were
reduced with the Gemini IRAF package,
2using dome flat fields
taken during the commissioning run.
Table 2, column (4) gives the predicted absolute throughput
of GMOS based on the transmission of the main optics, i.e.,
the collimator and camera lens groups, the filters, and the mean
quantum efficiency of the CCDs (see Fig. 2) at the central
wavelengths of the imaging filters. These response functions
were multiplied together with the telescope and atmosphere
response functions, also given in Table 2, and were used to
derive expected counts for a given standard star magnitude
(assuming spectral types, which were chosen based on the pub-
lished broadband colors from Landolt [1992]). The telescope
response function used in this calculation was derived from the
2
IRAF is distributed by National Optical Astronomy Observatories, which is operated by the Association of Universities for Research in Astronomy, Inc., (AURA), under cooperative agreement with the National Science Foundation. The Gemini IRAF package is distributed by Gemini Observatory, AURA.
Figure 3.1: Schematic diagram showing the layout of the three GMOS-N de- tectors. The small gaps between the detectors of about 0.5 mm correspond to about 37 pixels. The shaded region represents the imaging field of view while the dotted line shows the on-instrument wave-front sensor patrol field, projected onto the detector plane [150].
Full Unignetted Field Size 60 arc-minutes diameter
Minimum Fiber to Fiber Separation 37 arc-seconds
Positioning Accuracy 0.3 arc-seconds rms
Configuration Time (100 Fibers) 20-25 minutes
Total Number of Fiber Slots 288
Number of Guide Fibers 10
Number of Available Science Cables 2
Number of Active Fibers per Cable 90 Red, 83 Blue
Fiber Cable Length 25 meters
Blue Cable Spectral Window 3000-7000 Angstrom
Blue Cable Fiber Diameter 3.1 arc-seconds
Red Cable Spectral Window 4000 Angstrom- 1.8µm
Red Cable Fiber Diameter 2.0 arc-seconds
Table 3.3: Hydra Positioner Characteristics
3.2.3
IFU
GMOS-N is also equipped with an Integral Field Unit (IFU) [11] making it possible
to obtain spectra simultaneously of an area of about 35 arcsec2 with a sampling
of 0.2 arcsec. This mode is based upon using a 1500 element array in the pre-slit environment to slice the focal plane into a multitude of small components which allows to reconstruct an image at a particular wavelength, or extract a spectrum
from any point in the field of view. The science field of view is 35 arcsec2 and is
sampled by 1000 elements. The sky is sampled with 500 elements which are located ∼1 arcmin away from science field of view.
3.2.4
Nod & Shuffle
One of the most interesting modes on GMOS is Nod & Shuffle [124] which provides superior sky subtraction and increases the density of the slits. Its main handicaps are the increased observing time required and the smaller field of view. The basic concept of this method, which is adapted from the near-infrared astronomy nod- ding or beam-switching, is that unilluminated portions of the CCD can be used for storage. In this mode the telescope is frequently nodded between an object and a sky position while at the same time it shuffles the charge on the CCD detectors between science and unilluminated regions. This technique provides the observer with images that contain two spectra, one of the object and one of the sky. Even though these two spectra are stored in different regions of the CCD, they were im- aged with exactly the same pixels through identical optical paths. As a result when one subtracts the sky spectrum from the object spectrum one will be able to achieve
Figure 3.2: Illustration of the nod-shuffle procedure implemented in the LDSS spectrograph showing progressive stages of image formation. (a) The spectra of the objects through the slits is imaged onto the central portion of an oversized CCD. (b) The first image is shuffled up into a storage region (with the shutter closed), and the telescope is offset to adjacent sky which is then imaged onto the now empty central region of the detector. (c) The object image is shuffled back and additional object photons are imaged. (d) Sky is shuffled back and imaged. Steps (c) and (d) are cycled continuously until the integration is complete. Figure and caption from [124].
Available Gratings 7
Minimum Fiber to Fiber Separation 37 arc-seconds
Cameras Simmons (f.i. 381 mm, 3000 ˚A-1.5µm)
Bench (f.i. 285 mm, 3800 ˚A-1.5µm)
Total Number of Fiber Slots 288
Collimator 6 inch f/6.7 Paraboloid
Camera Collimator Angle Variable between 11 and 45 degrees
Detector 2048 (24µm-pixels) CCD
Typical Resolution Element 2 to 4 pixels
Spectral Coverage 100 ˚A to 1 octave
Table 3.4: Bench Spectrograph Characteristics
sky subtraction accuracies as good as 0.04% [124] with the trade-off of almost a
factor √2 higher noise in the sky-subtracted spectrum. Due to the fact that the sky
spectrum is derived from regions adjacent to the object much shorter slits can be used. This results in an increase of the number of slits in MOS mode of up to a factor of 10. However since a part of the detector is used for storage and remains unilluminated the field of view is reduced by up to 50%. An example of how Nod & Shuffle works is given in Figure 3.2.